[lammps-users] Si-O MEAM problems


I am trying to implement the MEAM potential for the Si-O-Au system described in [Kuo, Clancy, Modelling Simul. Mater. Sci. Eng vol12 (p1309-1329), 2005] in LAMMPS.
I tried to carefully put the values from the paper into the LAMMPS MEAM parameter files.

Bulk systems consisting of only one atom type seem to work fine with their values, but when I try to simulate a-quartz (Si-O) the system starts to behave in an unphysical manner (significant loss of volume). Similar problems were reported earlier on this forum no solution was found?

After investigation I found out atleast three differences in the formulas described in Gullet (a reference from the LAMMPS meam documentation page) and in Kuo-Clancy paper

  1. The Rose energy equation used in lammps MEAM deviates slightly from the one used in Kuo-Clancy. However, a similar equation to Kuo-Clancy was found commented in the lammps MEAM library code. I simply uncommented that line and commented the existing one as a solution.

  2. The third partial electron density (eqn 4.7d in Gullet) is lacking the 3/5* -term in Kuo-Clancy. This can be probably fixed by using the augt1 flag?

  3. Kuo-Clancy seem to use the t_i^k values as constants, whereas Gullet has a function for them (eqn 4.6 in Gullet). I don’t know how to fix this. Any help would be highly appreciated.

I am using the LAMMPS (15 Jan 2010) distribution.

Aleksi Leino

I think you should send these Qs to Greg Wagner who
did the MEAM implementation in LAMMPS (gjwagn at sandia.gov).
You might also try the most current LAMMPS - Greg has a made
a couple small changes to the MEAM files in the last month or so.